Biomarker for lung cancer diagnosis and artificial intelligence-based information providing method for lung cancer diagnosis

Abstract

The present invention relates to a biomarker for lung cancer diagnosis and an artificial intelligence-based information providing method for lung cancer diagnosis. More specifically, it relates to a biomarker composition for lung cancer diagnosis containing kynurenine (KN), lysophosphatidylcholine 16:0 (LPC16), lysophosphatidylcholine 18:0 (LPC18), glutamic acid (Glu), hexanoyl-L-carnitine (HC), octanoyl-L-carnitine (OC), decanoyl-L-carnitine (DC), dodecanoyl-L-carnitine (DDC), myristoyl-L-carnitine (MC), and palmitoyl-L-carnitine (PC), and an information providing method for artificial intelligence-based lung cancer diagnosis using the above biomarkers. As a result of establishing an artificial intelligence-based algorithm model for lung cancer diagnosis using the biomarkers for lung cancer diagnosis selected in the present invention, the early lung cancer screening ability has a sensitivity of 90.21% and a specificity of 93.03%, which is confirmed to have very high accuracy compared to conventional lung cancer screening methods. Therefore, the present invention can effectively provide information related to lung cancer diagnosis.

Description

Technical Field

[0001] The present invention relates to a biomarker for lung cancer diagnosis and an artificial intelligence-based information providing method for lung cancer diagnosis. More specifically, it relates to a biomarker composition for lung cancer diagnosis containing kynurenine (KN), lysophosphatidylcholine 16:0 (LPC16), lysophosphatidylcholine 18:0 (LPC18), glutamic acid (Glu), hexanoyl-L-carnitine (HC), octanoyl-L-carnitine (OC), decanoyl-L-carnitine (DC), dodecanoyl-L-carnitine (DDC), myristoyl-L-carnitine (MC), and palmitoyl-L-carnitine (PC), and an artificial intelligence-based information providing method for lung cancer diagnosis using the above biomarkers.

Background Art

[0002] Cancer is a disease in which cells proliferate infinitely and interfere with the functions of normal cells. Representative examples include liver cancer, lung cancer, gastric cancer, breast cancer, colorectal cancer, and ovarian cancer. Substantially, cancer can occur in any tissue. Early cancer diagnosis was based on external changes in biological tissues associated with the growth of cancer cells. In recent years, diagnoses using the detection of trace biomolecules present in biological tissues or cells such as blood, glycochains, and DNA have been attempted. However, the most commonly used cancer diagnosis methods are those using tissue samples obtained through biopsy or diagnosis using images.

[0003] Among them, a biopsy has the disadvantages of causing great pain to the patient, being costly, and taking a long time to diagnose. In addition, when a patient actually has cancer, there is a risk of inducing cancer metastasis during the biopsy process. In the case of a site where a tissue sample cannot be obtained through a biopsy, the disease cannot be diagnosed until the suspected tissue is removed surgically. In particular, lung cancer is one of the cancers with a high global fatality rate. Currently, in the case of lung cancer, there is a high dependence on imaging methods (such as X-rays, CTs, MRIs, etc.). However, more than half of lung cancer patients are already inoperable at the time of detection. Even if they are judged to be operable and surgery is performed, in many cases, complete resection is impossible. Therefore, in order to increase the cure rate of lung cancer, early diagnosis and treatment of lung cancer are of utmost importance. However, since there are limited diagnostic markers useful for lung cancer, there is a problem that such diagnosis is difficult. Therefore, it is necessary to search for cancer-specific markers present in biological samples and develop a method that can diagnose cancer with high accuracy and precision using these markers. Recently, methods using artificial intelligence have been studied for more accurate cancer diagnosis. Machine learning or deep learning is mainly used for artificial intelligence-based analysis, and various studies are being conducted to apply such artificial intelligence to the bio field (Korean Patent Publication No. 10-2014-0002149, Korean Registered Patent No. 10-2268963).

Summary of the Invention

Problems to be Solved by the Invention

[0004] Therefore, as a result of intensive efforts to select markers that can more accurately diagnose lung cancer, the present inventors have selected biomarkers including kynurenine (KN), lysophosphatidylcholine 16:0 (LPC16), lysophosphatidylcholine 18:0 (LPC18), glutamic acid (Glu), hexanoyl-L-carnitine (HC), octanoyl-L-carnitine (OC), decanoyl-L-carnitine (DC), dodecanoyl-L-carnitine (DDC), myristoyl-L-carnitine (MC), and palmitoyl-L-carnitine (PC), and confirmed that the expression levels of these biomarkers show different patterns between the patient group and the normal control group. In addition, as a result of analyzing the quantitative values for these 10 biomarkers using an algorithm based on artificial intelligence, it was confirmed that the diagnostic ability for lung cancer was improved, and the present invention was completed.

[0005] Accordingly, an object of the present invention is to provide a biomarker composition for diagnosing lung cancer. Another object of the present invention is to provide a composition for diagnosing lung cancer, which contains a substance for measuring the expression level of the above biomarker, and a lung cancer diagnostic kit using the same.

[0006] Another object of the present invention is to provide a method for providing artificial intelligence-based information for diagnosing lung cancer using the above biomarker.

Means for Solving the Problems

[0007] To achieve the above object, The present invention provides a biomarker composition for lung cancer diagnosis, which contains kynurenine (KN), lysophosphatidylcholine 16:0 (LPC16), lysophosphatidylcholine 18:0 (LPC18), glutamic acid (Glu), hexanoyl-L-carnitine (HC), octanoyl-L-carnitine (OC), decanoyl-L-carnitine (DC), dodecanoyl-L-carnitine (DDC), myristoyl-L-carnitine (MC), and palmitoyl-L-carnitine (PC).

[0008] In a preferred embodiment of the present invention, the above-mentioned biomarker can be extracted from blood. In another preferred embodiment of the present invention, the above-mentioned blood may be whole blood, plasma or serum. To achieve other purposes, The present invention provides a composition for lung cancer diagnosis, which contains a preparation for measuring the blood concentration of the above-mentioned biomarker composition for lung cancer diagnosis.

[0009] In a preferred embodiment of the present invention, the preparation for measuring the level of the above-mentioned biomarker composition may be a preparation for mass spectrometry. Furthermore, the present invention provides a kit for lung cancer diagnosis, which contains a preparation for measuring the blood concentration of the above-mentioned biomarker composition for lung cancer diagnosis.

[0010] To achieve another purpose, The present invention includes: (a) measuring the levels of biomarkers for lung cancer diagnosis, including kynurenine (KN), lysophosphatidylcholine 16:0 (LPC16), lysophosphatidylcholine 18:0 (LPC18), glutamic acid (Glu), hexanoyl-L-carnitine (HC), octanoyl-L-carnitine (OC), decanoyl-L-carnitine (DC), dodecanoyl-L-carnitine (DDC), myristoyl-L-carnitine (MC), and palmitoyl-L-carnitine (PC) from the blood of a subject; and (b) applying the expression levels of the above biomarkers to a machine learning algorithm model; and provides a method for providing information for lung cancer diagnosis using artificial intelligence.

[0011] In a preferred embodiment of the present invention, the blood in step (a) may be whole blood, plasma or serum. In another preferred embodiment of the present invention, the concentration of the biomarker in step (a) can be obtained by mass analyzing whole blood, plasma or serum samples by mass peak area. Specifically, it can be obtained through liquid chromatography-mass spectrometry (LC-MS). The mass spectrometer may be any one of Triple TOF, Triple Quadrupole, and MALDI TOF capable of quantitative measurement. In another preferred embodiment of the present invention, in the step of applying to the above algorithm model in (b), the biomarker level in the blood of the subject to be examined can be input into the above algorithm model, and the presence or absence of lung cancer can be output as an output value.

[0012] In another preferred embodiment of the present invention, the algorithm model in the above step (b) is (i) measuring the levels of biomarkers for lung cancer diagnosis, including kynurenine (KN), lysophosphatidylcholine 16:0 (LPC16), lysophosphatidylcholine 18:0 (LPC18), glutamic acid (Glu), hexanoyl-L-carnitine (HC), octanoyl-L-carnitine (OC), decanoyl-L-carnitine (DC), dodecanoyl-L-carnitine (DDC), myristoyl-L-carnitine (MC), and palmitoyl-L-carnitine (PC) from the blood of lung cancer patients and normal control groups; and (ii) deriving through the step of training the levels of the above biomarkers with a machine learning algorithm to generate a lung cancer onset prediction model.

[0013] In another preferred embodiment of the present invention, the above artificial intelligence can be machine learning or deep learning. More specifically, the algorithm in the above step (b) is any one selected from linear or non-linear classification algorithms including k-nearest neighbor algorithm; logistic regression algorithm; discriminant analysis algorithm; partial least squares-discriminant analysis algorithm; support vector machine algorithm; decision tree algorithm; decision tree ensemble algorithm; and neural network algorithm. In another preferred example of the present invention, when the above algorithm is a support vector machine algorithm, it can be represented by the kernel function of the following formula 1.

[0014] [Number]

[0015] The present invention also provides an artificial intelligence-based lung cancer diagnosis prediction device including a measurement unit that measures the level of a biomarker for lung cancer diagnosis in the blood of a subject to be examined; and a cancer diagnosis unit that inputs the level of the biomarker into a learned artificial intelligence algorithm to determine the presence or absence of lung cancer onset. [Advantages of the Invention]

[0016] In the present invention, ten biomarkers for more accurately diagnosing lung cancer were selected, and an artificial intelligence-based algorithm for lung cancer diagnosis was established using these biomarkers. The screening ability for early lung cancer using the algorithm developed in the present invention has a sensitivity of 90.21% and a specificity of 93.03%, and it was confirmed that it has very high accuracy compared to existing lung cancer screening methods. Therefore, the present invention can effectively provide information regarding lung cancer diagnosis. [Brief Description of the Drawings]

[0017]

Figure 1

Figure 2

Modes for Carrying Out the Invention

[0018] The present invention will be described in detail below.

[0019] <Biomarker Composition for Lung Cancer Diagnosis> From one aspect, the present invention relates to a biomarker composition for lung cancer diagnosis, which contains kynurenine (KN), lysophosphatidylcholine 16:0 (LPC16), lysophosphatidylcholine 18:0 (LPC18), glutamic acid (Glu), hexanoyl-L-carnitine (HC), octanoyl-L-carnitine (OC), decanoyl-L-carnitine (DC), dodecanoyl-L-carnitine (DDC), myristoyl-L-carnitine (MC), and palmitoyl-L-carnitine (PC).

[0020] In the present invention, the above-mentioned biomarker is extracted from blood, and the blood may be whole blood, plasma or serum. The term "diagnosis" used in the present invention means to confirm the existence or characteristics of a pathological condition. For the purpose of the present invention, the diagnosis is to confirm the presence or absence of the onset of lung cancer.

[0021] The term "diagnostic biomarker" used in the present invention means an organic biomolecule such as a polypeptide, nucleic acid (e.g., mRNA, etc.), lipid, glycolipid, glycoprotein, sugar (monosaccharide, disaccharide, oligosaccharide, etc.) that shows a significant increase or decrease in the lung cancer patient group compared with the normal control group, and preferably the above-mentioned biomarker composition for lung cancer diagnosis. In a specific embodiment of the present invention, blood was collected from the normal control group (Con) and the lung cancer (LC) patient group, and the concentrations of KN, LPC16, LPC18, Glu, HC, HC, OC, DC, DDC, MC, and PC in the blood were measured. It was confirmed that the blood concentrations of the biomarkers of the present invention were significantly different between the quantitative values of the lung cancer patient group and the normal control group. It was confirmed that the concentrations of KN, LPC16, LPC18, and Glu metabolites increased in the blood of the lung cancer patient group compared to the normal control group, while the concentrations of HC, OC, DC, DDC, MC, and PC metabolites decreased (FIG. 1 and FIG. 2).

[0022] <Composition for lung cancer diagnosis> From another perspective, the present invention relates to a composition for lung cancer diagnosis containing a preparation for measuring the blood concentration of the biomarker composition for lung cancer diagnosis of the present invention. The composition for lung cancer diagnosis according to the present invention shall apply mutatis mutandis to the above <biomarker composition for lung cancer diagnosis>. The biomarker of the present invention is a metabolite, and the preparation for measuring the level of the above biomarker composition may be a preparation for mass spectrometry. The above preparation for mass spectrometry is a preparation capable of analyzing the mass of the marker in whole blood, plasma, or serum, and specifically means a preparation capable of performing liquid chromatography-mass spectrometry (LC-MS).

[0023] <Kit for lung cancer diagnosis> In another aspect, the present invention relates to a kit for lung cancer diagnosis containing a preparation for measuring the blood concentration of the biomarker composition for lung cancer diagnosis of the present invention.

[0024] The composition for lung cancer diagnosis according to the present invention shall apply mutatis mutandis to the above <biomarker composition for lung cancer diagnosis>. The above kit can be manufactured by a conventional manufacturing method known in the art. The above kit can include, for example, an antibody in lyophilized form, a buffer solution, a stabilizer, an inert protein, etc.

[0025] <Method for providing information on lung cancer diagnosis using an artificial intelligence-based algorithm> In another aspect, the present invention provides a method for providing information for diagnosing lung cancer using artificial intelligence, comprising: (a) measuring the levels of biomarkers for diagnosing lung cancer, including kynurenine (KN), lysophosphatidylcholine 16:0 (LPC16), lysophosphatidylcholine 18:0 (LPC18), glutamic acid (Glu), hexanoyl-L-carnitine (HC), octanoyl-L-carnitine (OC), decanoyl-L-carnitine (DC), dodecanoyl-L-carnitine (DDC), myristoyl-L-carnitine (MC), and palmitoyl-L-carnitine (PC) from the blood of a subject; and (b) applying the expression levels of the above biomarkers to a machine learning algorithm model.

[0026] In the present invention, the blood in step (a) may be whole blood, plasma or serum. In the present invention, the concentration of the biomarker in step (a) can be obtained by mass spectrometry of whole blood, plasma or serum samples using mass peak area. Specifically, it can be obtained by liquid chromatography-mass spectrometer (LC-MS). The mass spectrometer may be any one of Triple TOF, Triple Quadrupole, and MALDI TOF capable of quantitative measurement.

[0027] In the present invention, the step of applying to the algorithm model in (b) can input the level of the biomarker in the blood of the subject to be tested into the algorithm model and output the presence or absence of lung cancer onset as an output value. In the present invention, the algorithm model in step (b) is (i) Measuring the levels of biomarkers for lung cancer diagnosis, including kynurenine (KN), lysophosphatidylcholine 16:0 (LPC16), lysophosphatidylcholine 18:0 (LPC18), glutamic acid (Glu), hexanoyl-L-carnitine (HC), octanoyl-L-carnitine (OC), decanoyl-L-carnitine (DC), dodecanoyl-L-carnitine (DDC), myristoyl-L-carnitine (MC), and palmitoyl-L-carnitine (PC) from the blood of lung cancer patients and normal control groups; and (ii) Derivable through the steps of training the levels of the above biomarkers with a machine learning algorithm to generate a lung cancer onset prediction model.

[0028] In the present invention, the above artificial intelligence can be machine learning or deep learning. More specifically, the algorithm in step (b) can be any one selected from linear or non-linear classification algorithms including k-nearest neighbor algorithm; logistic regression algorithm; discriminant analysis algorithm; partial least squares-discriminant analysis algorithm; support vector machine algorithm; decision tree algorithm; decision tree ensemble algorithm; and neural network algorithm.

[0029] In the present invention, when the above algorithm is a support vector machine algorithm, it can be represented by the kernel function of the following Equation 1:

[0030]

Equation

[0031] In a specific embodiment of the present invention, a prediction model was developed using the quantitative values of the 10 biomarkers of the present invention measured in the blood of a lung cancer patient group and a normal control group, and the algorithm utilized a support vector machine with a radial basis function as the kernel. As a result of confirming the diagnostic ability of early-stage lung cancer using the developed prediction model, it was found that the accuracy showed a sensitivity of 90.21% and a specificity of 93.03%. That is, it was confirmed that the artificial intelligence-based lung cancer diagnosis method using the 10 biomarkers of the present invention has very high accuracy compared to other existing diagnostic methods.

[0032] In another aspect, the present invention also provides a measurement unit for measuring the levels of lung cancer diagnostic biomarkers including kynurenine (KN), lysophosphatidylcholine 16:0 (LPC16), lysophosphatidylcholine 18:0 (LPC18), glutamic acid (Glu), hexanoyl-L-carnitine (HC), octanoyl-L-carnitine (OC), decanoyl-L-carnitine (DC), dodecanoyl-L-carnitine (DDC), myristoyl-L-carnitine (MC), and palmitoyl-L-carnitine (PC) in the blood of a subject; and a cancer diagnosis unit that inputs the above biomarker levels into a learned artificial intelligence algorithm to determine the presence or absence of lung cancer; and provides an artificial intelligence-based lung cancer diagnosis prediction device.

[0033] The present invention will be further described in detail through the following examples. It will be apparent to those skilled in the art that these examples are merely for illustrative purposes of the present invention and should not be construed as limiting the scope of the present invention.

[0034] <Example 1: Measurement of Biomarker Concentrations in the Blood of Lung Cancer Patients and Normal Control Groups> [1-1: Preparation of Samples] To confirm whether the biomarker of the present invention can diagnose lung cancer, a total of 864 people were screened. Specifically, 445 people in the normal control group and 419 people in the lung cancer (LC) patient group were screened through Bundang Seoul National University Hospital and Asia University Hospital, and blood was collected.

[0035] [1-2: Measurement of Biomarker Concentrations in the Blood] Plasma was separated from the above blood samples, and standard substances for each biomarker at different concentrations required for the biomarker standard curves of kynurenine (KN), lysophosphatidylcholine 16:0 (LPC16), lysophosphatidylcholine 18:0 (LPC18), glutamic acid (Glu), hexanoyl-L-carnitine (HC), octanoyl-L-carnitine (OC), decanoyl-L-carnitine (DC), dodecanoyl-L-carnitine (DDC), myristoyl-L-carnitine (MC), and palmitoyl-L-carnitine (PC) were prepared.

[0036] 400 μl of lipid extraction buffer (Abnova, Taiwan) was added to 20 μl of plasma and standards. After vortexing, centrifugation (10,000 g, 5 min, 4 °C) was performed. After centrifugation, all of the supernatant was transferred to a new tube and dried using a concentrator for 12 - 16 h. 50 μl of 100% methanol containing 0.1% formic acid (FA) was added to the dried metabolite extract, and after thoroughly dissolving using a vortex, analysis was performed using LC - MS / MS (liquid chromatography - mass spectrometry / mass spectrometry).

[0037] The LC used was a Shimadzu LC 40 system, and the MS was an AB Sciex Triple Quad 5500+ system. The MS was equipped with a turbo spray ion source. The analytical samples were separated within a BEH C18 column (1.7 μm, 2.1 * 50 mm; Waters) of the Shimadzu LC 40 system. The solvent used was a two - step linear gradient (solvent A, 0.1% FA in water; solvent B, 0.1% FA in 100% acetonitrile; 5 - 55% solvent B for 2.5 min, 55% solvent B for 5.5 min, 55 - 95% solvent B for 7.5 min, 95% solvent B for 11 min, 95 - 5% solvent B for 11.1 min, and 5% solvent B for 14.5 min).

[0038] Mass spectrometry (MS / MS) was performed using the MRM (multiple reaction monitoring) mode. Among the mass spectra in the same time zone as the time zone when the metabolites corresponding to each biomarker passed through liquid chromatography, the areas of the mass peaks with the same mass value were calculated. Standard curves were created using the mass peaks of each biomarker standard substance, and the concentrations of each biomarker were measured by substituting the mass peaks of each sample into the standard curves.

[0039] As a result, as shown in FIGS. 1 and 2, it was confirmed that there were significant differences in the blood concentrations of 10 biomarkers between the quantitative values of the lung cancer patient group and the quantitative values of the normal control group. Compared with the normal control group, the blood concentrations of KN, LPC16, LPC18, and Glu metabolites increased in the lung cancer patient group, and the blood concentrations of HC, OC, DC, DDC, MC, and PC metabolites decreased.

[0040] <Example 2: Development of an Artificial Intelligence-Based Algorithm Model for Lung Cancer Diagnosis> In the present invention, a support vector machine algorithm using a radial basis function as a kernel was applied to the quantitative values for 10 biomarkers, and a prediction model capable of diagnosing the presence or absence of lung cancer was developed.

[0041] Using the kernel function represented by the following formula 1, the lung cancer onset prediction model was trained through tuning of algorithm parameters.

[0042]

Equation

[0043] The parameter σ in Equation 1 determines the range of influence given by one training sample, and another parameter C determines to what extent the misclassification of training samples is tolerated. Since both parameters cause the learning model to be underfitting or overfitting depending on their values, the optimal parameters were selected through iterative cross-validation.

[0044]

Table 1

[0045] As a result of confirming the diagnostic ability of early lung cancer using the developed prediction model, as shown in Table 1 above, it was found that the accuracy was 90.21% in sensitivity, 93.03% in specificity, 92.42% in positive predictive value (PPV), and 90.99% in negative predictive value (NPV). That is, it was confirmed that the artificial intelligence-based lung cancer diagnosis method using the 10 biomarkers of the present invention has very high accuracy compared to other existing diagnostic methods.

[0046] In the present invention, 10 biomarkers that can more accurately diagnose lung cancer were selected, and the early lung cancer screening ability using the algorithm developed using these biomarkers was 90.21% in sensitivity and 93.03% in specificity. Since it was confirmed that it has very high accuracy compared to existing lung cancer screening methods, the present invention can be effectively applied to lung cancer diagnosis.

Claims

1. A biomarker composition for lung cancer diagnosis, comprising kynurenine (KN), lysophosphatidylcholine 16:0 (LPC16), lysophosphatidylcholine 18:0 (LPC18), glutamic acid (Glu), hexanoyl-L-carnitine (HC), octanoyl-L-carnitine (OC), decanoyl-L-carnitine (DC), dodecanoyl-L-carnitine (DDC), myristoyl-L-carnitine (MC), and palmitoyl-L-carnitine (PC).

2. The biomarker composition for lung cancer diagnosis according to claim 1, wherein the biomarker is extracted from blood.

3. The biomarker composition for lung cancer diagnosis according to claim 1, wherein the blood is whole blood, plasma or serum.

4. A composition for lung cancer diagnosis, comprising a preparation for measuring the blood concentration of the biomarker composition for lung cancer diagnosis according to any one of claims 1 to 3.

5. The composition for lung cancer diagnosis according to claim 4, wherein the preparation for measuring the level of the biomarker composition is a preparation for mass spectrometry.

6. A kit for lung cancer diagnosis, comprising a preparation for measuring the blood concentration of the biomarker composition for lung cancer diagnosis according to any one of claims 1 to 3.

7. (a) Measuring the levels of biomarkers for lung cancer diagnosis, including kynurenine (KN), lysophosphatidylcholine 16:0 (LPC16), lysophosphatidylcholine 18:0 (LPC18), glutamic acid (Glu), hexanoyl-L-carnitine (HC), octanoyl-L-carnitine (OC), decanoyl-L-carnitine (DC), dodecanoyl-L-carnitine (DDC), myristoyl-L-carnitine (MC), and palmitoyl-L-carnitine (PC) from the blood of a subject; and (b) Applying the expression levels of the above biomarkers to a machine learning algorithm model; A method for providing information for lung cancer diagnosis using artificial intelligence, comprising the steps of: (Claim 8) (3) The method for providing information for lung cancer diagnosis using artificial intelligence according to claim 7, wherein the blood in step (a) is whole blood, plasma or serum. (Claim 9) (5) The method for providing information for lung cancer diagnosis using artificial intelligence according to claim 7, wherein in step (a), the measurement of the biomarker level is obtained by liquid chromatography-mass spectrometry (LC-MS). (Claim 10) (7) The step of applying to the algorithm model in (b) is to input the level of the biomarker in the blood of the subject to be examined into the algorithm model and output the presence or absence of lung cancer onset as an output value. A method for providing information for lung cancer diagnosis using artificial intelligence according to claim 7. (Claim 11) (9) The algorithm model in step (b) is (i)Measuring the levels of lung cancer diagnostic biomarkers including kynurenine (KN), lysophosphatidylcholine 16:0 (LPC16), lysophosphatidylcholine 18:0 (LPC18), glutamic acid (Glu), hexanoyl-L-carnitine (HC), octanoyl-L-carnitine (OC), decanoyl-L-carnitine (DC), dodecanoyl-L-carnitine (DDC), myristoyl-L-carnitine (MC), and palmitoyl-L-carnitine (PC) from the blood of lung cancer patients and normal control groups; (ii) Deriving through the steps of training the levels of the above biomarkers with a machine learning algorithm to generate a lung cancer onset prediction model. A method for providing information for lung cancer diagnosis using artificial intelligence according to claim 7, characterized in that it is derived through these steps.

12. The algorithm in the above step (b) is selected from linear or non-linear classification algorithms including k-nearest neighbor algorithm; logistic regression algorithm; discriminant analysis algorithm; partial least squares-discriminant analysis algorithm; support vector machine algorithm; decision tree algorithm; decision tree ensemble algorithm; and neural network algorithm. A method for providing information for lung cancer diagnosis using artificial intelligence according to claim 7, characterized in that it is so selected.

13. When the algorithm in the above step (b) is a support vector machine algorithm, it is characterized in that it is represented by the kernel function of the following formula 1. A method for providing information for lung cancer diagnosis using artificial intelligence according to claim 7: 【Number 1】 In the above formula 1, x is the measured blood level value of the lung cancer diagnostic biomarker composition, and σ is the parameter of the flexibility (curvature) of the crystal boundary.

14. A measurement unit that measures the levels of biomarkers for lung cancer diagnosis, including kynurenine (Kyn), lysophosphatidylcholine 16:0 (LPC16), lysophosphatidylcholine 18:0 (LPC18), glutamic acid (Glu), hexanoyl-L-carnitine (HC), octanoyl-L-carnitine (OC), decanoyl-L-carnitine (DC), dodecanoyl-L-carnitine (DDC), myristoyl-L-carnitine (MC), and palmitoyl-L-carnitine (PC) in the blood of the subject to be examined; and A cancer diagnosis unit that inputs the levels of the above biomarkers into a learned artificial intelligence algorithm to determine the presence or absence of lung cancer onset; An artificial intelligence-based lung cancer diagnosis prediction device comprising.

Images